Predictors of unplanned readmission to acute care from inpatient brain injury rehabilitation.

Abstract

To identify the predictors of unplanned readmission to acute care (RTAC) from inpatient brain injury rehabilitation and to develop a risk prediction model. RTAC from inpatient rehabilitation is not uncommon. Individual rehabilitation patient populations require their own body of evidence regarding predictors of RTAC. Retrospective cohort study. Adult patients with new onset acquired brain injury admitted to a stand-alone rehabilitation facility between 1 January 2012-31 December 2018 were included in the study. The main measures were RTAC, sensitivity, specificity, the C-statistic and Youden's index. This paper is reported using the Strengthening the Reporting of Observational Studies in Epidemiology guidelines. Of 383 patients admitted for rehabilitation, 83 (22%) experienced a RTAC; 69 (18%) patients had at least one unplanned RTAC episode. Patients requiring unplanned RTAC were more likely to have lower Glasgow Coma Scale (GCS) and Functional Independence Measure (FIM) scores on rehabilitation admission, a higher burden of care on rehabilitation discharge and be discharged to a nonhome residence. Rehabilitation admission GCS and motor FIM were identified as the independent RTAC predictors in multivariate regression modelling. The combined C-statistic was 0.86. A GCS cut-off score of≤14 and motor FIM cut-off score of≤40 were identified as optimal, yielding a combined Youden's index of 0.56 (sensitivity=0.72; specificity=0.83). Patients requiring an unplanned RTAC had a lower functional status on rehabilitation admission. A prediction model for unplanned RTAC has been developed using validated and readily available clinical measures. The developed RTAC risk prediction model is the first step in preventing unplanned RTAC from inpatient brain injury rehabilitation. Future research should focus on discrete interventions for preventing unplanned RTAC from inpatient brain injury rehabilitation.